Multi-ant colony optimization based on bidirectional induction mechanism and cooperative game

To address the lack of convergence speed and diversity of traditional ant colony algorithm when solving large-scale travelling salesmen problem (TSP), a multi-ant colony optimization based on bidirectional induction mechanism and cooperative game is proposed. First, a novel heterogeneous multi-population is introduced, different populations cooperate to enhance the algorithm's overall performance. Second, a cooperative game in multi-population is proposed, the pheromone income is distributed to participants based on their contribution, which can balance the convergence and diversity. Third, we introduce a bidirectional induction mechanism, by expanding the difference of pheromone concentration between outstanding solutions and mediocre solutions, the convergence can be further improved. In addition, the current optimal clip will be recommended to other populations for learning, each population will continue search based on the current optimal clip, which can improve the accuracy of the solutions. If the algorithm falls into stagnation, the current optimal solution will be inducted reversely, which can help the algorithm jump out of the local optimum. Finally, the experimental results of the large-scale TSP instances show that the improved algorithm can effectively balance the convergence and diversity, and it has better stability.